Econometric Replication explanation of Hansen, M. E., & Ziebarth, N. L. (2017). Credit relationships and business bankruptcy during the great depression. American Economic Journal: Macroeconomics, 9(2), 228‐55 (Only replicating Table 4)
Used the program Stata to recreate the findings of Table 4 in Hansen, M. E., & Ziebarth, N. L. (2017). Credit relationships and business bankruptcy during the great depression. American Economic Journal: Macroeconomics, 9(2), 228‐55
please use the pdf to guide through answering the following requirements in 2-3 pages.
How do your results compare to the original? Include a table with the authors’ results and yours and discuss the similarities and differences of the key results.
b. How successful were you at replicating the authors’ results? Explain how hard or easy it was and what additional steps if any did you have to take to replicate the results. If you were not able to replicate why not? Keep track of what you do to replicate.
c. How would you evaluate the authors’ replication files? Discuss their data, code, and documentation.
d. Why do you think your results matched or didn’t match?
Steps for replication
- Table 4: Differences in the probability of business exit in a specific time period
- We found that it was easy to replicate because the do file was very organized and they explained where the data came from as well as which table they correspond to. The do-file was also broken up into sections. The readme file was not that help getting started though because the only information it really had was to follow the steps on the do-file and said it was self-explanatory
3 Replication 1 data was used to compute exit rates (the data that was used in Table 4 of the paper) which is what was used to compute exit rates because it included the data from the D&B books (a credit rating agency)
4 Had to save the data on to our desktop by changing the location
5 Delete all information under table 4 → not needed, decided to keep all information above it because it defined important variables used in our regression
6 Esttab was used to create the regression tables and it was noted that we had to download it in the do-file
No other steps were needed but this was expected because in the author's readme file the results said they were self-explanatory which they were
-Results turned out to be identical to the original results found by Hansen and Ziebarth in Table 4.
-There was the same level of standard error was found
-The author provided a very organized and detailed do-file which made it very easy for us to replicate because as we noted before, we did not have to make any changes to the authors original do-file
Interpretation:
The table is showing the likelihood that a business or firm between the years 1930-1931 and 1931-1935 are likely to “exit” or discontinue business.
Column 1: a baseline linear probability model with no fixed effects beyond those required for a difference-in-differences specification.
Column 2: column 1 plus industry-specific time trends
Column 3: controls for credit rating
Column 4: doesn’t include business in the “Delta Region” but otherwise same controls as column 3
Long term financial distress is in columns 5-8 (1931-1935)
*Baseline: Exit rate is about 7% pts lower in St. Louis than Atlanta region
*Exit Rate is about 15 % pts higher in St. Louis than Atlanta region
*Exit rate is about 12 % points > in the St. Louis Fed-controlled part of Mississippi during the 1931-1935 period post-crisis ( ~ 3.5 percent annually).